3: The Current and Future Environment: An OverallAssessement

Donald J. KesslerNASA, Johnson Spa.c (enl.r, Houslon.7p.tu.\

SUMMARY

Orbital debris is of a concem in primarily two regions of Earthorbital space: low Eatth orbit and geosynchronous orbit. Thchazard to spacecraf't from orbital debris in low Eanh orbit hasalready exceeded the hazard fiom natural meteoroids. Thiswas predicted by models published over l0 years ago, and hadbeen verilied by measurements over tbe last l'ew years. Thesesame models also predict that cenain altitudes are al. or near a'tritical density," where the debris haeard will increase a-s aresult of random collision breakups. independent of futurespacecraft operational practices. Consequently. there is a needto make immediate changes in operational practices.

The cunent hazard in geosynchronous orbit has not likelyexceeded the hazard from meteoroids. However. models andmeasurcrnents of Ihe environment in geosynchronous orbit areinadequate; therefore there is currently not an adequate long-term environment management strategy for geosynchronous

orbit. A lolg{erm strategy is required because of the increas-ing use of geosynchronous orbit plus the lact that objectsremain in orbit essentially for€ver at this altitude. There is aneed lo understand various slrategy options before making sig-niticant operational changes.

INTRODUCTION

The unlimired bounds of space could lead one ro conclude thatwe would be incapable ofcausing an environrn€ntal issue in thisnew frcntier. This may be the clLse for most of space; however.Eanh orbital space is linite, and past spacecrafl operational prd.-tices have already produced an orbital debris envircnment thatwill likely alfect the design of mosr turure spacecrafl operatingin near-Earth orbital space. lf letl unchecked, this €nvironmentcould increase within the next century to the point thal someoperations either become too expensive or too risky, ln order toeffectively manage the environment, we need to understand theenvimnmert which we have already produced and the porcndalsourc-es fbr future orbital debris. With such an undentanding,we may prer€rve near-Eaih space for future generations with-out significantly altering the current planned acriviries in space.

There ar€ two major regions of Earth orbit where orbitaldebris is ofconcem: l. l,ow Eanh Orbit (LEO), usually thoughrof as being below 2000 km altitude. 2. Ceosynchronous orbir(CEO), at an altitude of about 35,8m km. The orbi(al debrisissues and solutions in thes€ iwo regions require differentapproaches, so it is best to discuss tjrcm separately. Thereforethis paper will be divided in two pans.

PART I - LOW EARTH ORBIT

A comparison of the hazards caused by ortital debris and nat-ural meteoroids provides a threshold by which levels of qon-

cem can b€ measured. [n low Eanh orbit, this comparison isfairly straight forward, and provides some insight to orbitaldebris issues. Therefore, il is desirable to lilst und€rstand themeteoroid environment.

Meteoroids

Meteoroids are part of the interplanetary environment andresult from the disinregration and fiagmentation ofcorn€ts andasteroids which orbit the sun. Meteoroids pass through Eanhorbital space, rader than orbit the Eanh, with a velocity distri-bution averaging abour 16 kn/s€c (Kessler, 1969). At any oneinstant, about 200 kg of meteoroid mass is within 2000 km ofth€ Eanh's surface. The largest fraction ofthis mass is in mete-oroids with diamerers of about 0.1 mm. A lesser fraction of themass is between me!€oroid diameters of I mm and I cm...thesiz€ interval responsible for the "l'alling stills" or meteo6observed at night. This distribution of mass and velocity is sut-ficient to require shielding on some spacccraft, depending onthe spacecraft size and desired reliability. Figure I describesthe cumulative flux as a function of meteoroid diameter (Griin.et al., 1985). The average number of impacts on any surface iscalculated by multiplying this flux by the producr of time andaverage cross-sectional area of that surface exposed to theenvironment.

Protection against this environmena requires a backgroundin the field of hypervelq.'ity impacts. and can becone complexwhen unique materials and geometric propenies are taken intoconsideration. However, as a rule ofthumb. aluminum bumoer

20

shields can be construcled to have a total added aluminum

thickness equal to the meteoroid diametcr to bc protccted

aSainst: a single sheel of aluminum would have to b€ about 5

times more massive than an aluminunr bumDer. while s{)me''multi-shock" shielding protection tschniques are about half

as li8hl ar an aluminum bumpcr.

Historically. shielding as light as thermal insulation blankelsarc usually sutficient lbr protecring vulnerable area.,i. such aswiring bundles. and prcssurized containers. on small,unmanned sDacecrdli over th€ir lifetimc. This results from thelict. as can be concluded fnrm Figure l. that only meteoroidssmaller lhan I mm in dianreter are likely to hil these vulnem-

ble areas over th€ spacecraft liletime. Larger. longer duratiorr

arrd high reliability spacecraft. such as the planned SpaceStation Freedom, would require much morc protection againsr

meteoroids. Shielding weights totaling several lhousand kilGgrarns would be required lo protecl the vulnemble areas such

as the habitalion mdules and fuel slorage tanks. These higher

weights result from the lact, as also can be concluded from

Figure L that shiclding is rtr;uired to protect againsl mere-

omids of about 1).5 cm ovcr lhe hundreds of souare meters ()f

Space Station vulnerable arears in order to obtain aboul one

chance in l0 rhat these are&s will not be Denetraled over the

Space Stations planned 30 year lifetime (Christianscn, et al..r990).

Fundrmenlsls oforbital dcbris in Leo

Within lhe same 20(n km atnve the Earth are approximately

7(X)0 man-made orbiting objects which have b€en calalogedhy lhe US Space Command wilh a total mass o1- aboul

I,m0,fiX) kg. Many of thes€ objecls are in near polar orbit. so

that their velociries relative to one anothcr cafl be as high astwice their orbilal velocity, o. around 15 knr/sec. The averagecollision velocity between any panicular spacecraft orbiting in

near-Eanh ortrihl space and the cahloged objects are a func-

tion of that spacecraft's inclination and ranges from about l0

kn/*- lbr low inclination spacecruft. lo about 13 km/sec iorn€ar polar orbits. Because thcse vel(xitics are not too differentthal for meleoroids, a compamble amount of orbital debrismass in any particula. size inlerval will produce a comparablecollision probability, and the damage resulting f'rom a collisionwill be similar for meteoroids and orbital debris of af$ul thesame,, ize. Since. most ol thc orbit ing mass t\ in intact space-

crali or rockel bodies...objects several nreteni in diameter. thehazard crealed by rhese objects is more rhan 4 orders of magni-lude larger than the hazard from meteoroids whicb lue severalmeters in diame(er. However, this akrne is not necessarily sig-

nilicant since meteoroids of this sizc are not a problem forspacecrafl. Howevcr, when combined v,/ith other data, twoissues become ohvious which have txlth shon and lons termimplications to the envircnment:

| . lf only 0.Ol% of the orbiting mass. or less mass than in a

single aveaage spacecrafl werc convcned into a size dis-

tribution similar to th€ size distnbution ol meteoroids. it

Dtttutkl J. Kessltr

would create a hazard simiia. t() the hazard trom mcte

oroids. Therc have been ak)ut lU) satellite brcakups duc

k) explosions in Eanh orbit...more (han enough fragmen-

tation ma.ss to create such ir hazard. hut the degrcc ol the

hazard depends on rhe fngmcnt sizc dis(ributi(D resuh-

ing fnrm these hreakups. These breakups an: the major

conccm for short term orbitaldebns considerations.

2. Random hypervelocity coliisions will srrln hegin hr.()n-ven the orbit ing mass of satel l i les into a size distr ibul ion

that is not too different than thc mcteoroid sirc dirrrihu

tion. Hypervelocity laboratory tests indicate rhar a hyper

velocity collision between an average spacecrali a d ir

several kilogram fragment can bc expectcd to producc a

large number ol frugments in the I mm to I cm sizc inter,

val.

Figure 2 gives the ci l lculdted l lux (K€ssler, l98l B) ofcuta-

logued objccts as a function of altirude for an orbiting space-

crafl for l9li7. when solar activity was low. and fbr 1991.

when vrlar activiry was high. Note that the high sola-r acrivity

has increased the atmospheric densily and rcduced lhe flux

below 600 km...this has occurred during previous high solar

activity periods- Bv 1997, when solar activity is expecred ro bc

lower. the Rux belo* 6(n km should retum to tbout its 1987

values. Figure 2 is averaged over inclinationl the llux ldr

spacecratl with low inclinations will be slightly lowcr tby

atxrut l0%) lhan given in lhe figure. while some inclinarions(e.9., 80 degrEes and l(X) degrecs). will experience twice the

flux given in the l igure (Kessler, et al. . l9u9-A). The col l islon

rate fot small spacecrafl would be small against the catakrgued

population: bowever spacecraft larger than xLbout l(X) melers

in diameter begin lo have a significant probability ol c()llidingwith a cataloged obje(t. lt is for this renson rhar collision

avoidance maneuvent are planned li)r the Space Station. Just as

significanlly, the total area of all cataloged objects is largcr

than the area oi a l0O meter diameter spacecraft; consequenrly.

since collision avoidance between all 7{X)O catalogued objects

is impractical. there is a near cenainty that lwo catnlogtuedobjects will collide in the relalively oear fulure. Random colli-sions ire lhe major concem for long term orbitul debris consid-eraIit)ns.

E rl! predlc-tlons

The earliest orbital deb.is studies by NASA were nlosrly con,

cerned with calculating thc collision probabilities betweenobjects large cooulih to be catalogued by NORAD (Donahm.

1970: Bnxrks, et al. , 1975). Al l catalogued objects are largerthan l0 cm in diameter. Frugmcntatkln data from ground

explosions and hypervelocity tests gathcred by NASA LanSleyResearch Center suggested thal a much larger popularion oluncatalogued objects must exisl in Eanh orbit (Bess. 1975,.

NASA, Johnson Space Center (JSC) used rhe Langle), data topredict r iulure uncatalogued population fmm random colli-sions, even if such an uncatalogued population did not cur-

rently exist {Kcssler. et al., 1978). These predictions werc later

Thc currenl and future e vironnenl 21

t06

t0-

to'

103

€ 'o'

s ro'g -; r0-E! to-'a!

I ro''

? ro'3

t ''o'o

€ ,o'u

E ,0"to

1o'7

103

1o'9

rnetsoroils @ 5O0 km ,(C{un, Et al., 1985) /

to'toli o l

d;am6ler lcml

Fiturc l: Meleroid environmrnl at 500 km altitude.

atr[uds lkmltr'rgurc 2: Average flux resulting from US Spac€ Comnand cataloged population-

.t; 10 -

E

Eop . ,= ,rA

E

ood|

R . ^ . o

22

expaoded to include an estimate of the uncatalogued popula-tion in l9?E (Kcssler. l98l).

Until 1984, therc were no measurem€nrs of debris in orbir.othcr than what could be catalogcd by NORAD. hedictions ofsmaller debris had to be based on models which described theprocess of satellite breakups in terms of number and velocityof breakup fragments, and models which predicted the rate thatdebris was removed from the environment fhrough atmos-pheric drag. To predict futurc enviroorrents then, as now,assumptions had to be made conceming future rraffic intospace and the rate that future satellites will breakup. A mea-sure of the ability of these models to predict the futur€ env!rooment can b€ obtained by comparing these early model pre-dictions with today's rEasr[€rnents.

Figure 3 shows a published predicrion (Ke$sler, l98l),bas€d on 1978 data, when the cata.logue contained about 4500objects. The 1978 Debris "Observed" line represents the cata-logued popularioo betwecn 600 km and llm km: the"Con€ated" line is a model p.ediction of the 1978 environ-ment ba$ed on a comparison of the siz€ disribuiion mea.surcdfrom explosions on the grould and the size disribution ofcat-alogued o(biting fragments. Note that rhis analysis predictedthat th€re were about twice as many objects in orbit thal werelargc. than lO cm than is rcffected in the catalogue, and morethan three times as many objects larger thall 4 cm. The aaalysisalso predicted that by 1995 thcre would have been 3 satellitebrealups caused by random collisions, producing a populationof smaller debris that produced a debris flux which exceed therEteoroid flux for sizes larger than aboul I mm. Since debrisin these size ranges have now tren measurcd, and 1995 is lessthan 3 years away, these predictions can be comparcd withrecent measut€ments.

Measurcments of u[crtrlogu€d po]||drdot|The number of orbiting objects irlcreas€s with decreasing size.If one were to try to catalogue all orbiting objects, eventually,tb€ catalogue would becomc so largc that only a statisticalinterpretation of the population would be meaningful.Consequently, dEre is a debris size where statistical measure-ments become more cosl effective. Statistical measuremenissample a fraction of the population and do no( require that theeach detf.ted object be tracked so that it can b€ observedagain.

Sampling of objecrs in Earrh ortrit is a less difticulr problemthsr tlle cataloging of objects: however. the detection of uncaFalogued objecb rEquires either different senson, or rhat thesesensors be operatcd io a diflercnt mode than the senson used tocatalogu€. Remote sensoni are required for debris larger thanabout I mm, simply because the population ofthis size debris issufliciently spa$e that a large collection area is required inorder to obtain a shrisdcally meaningfut sample. Below I mm,the population is sufficiently dense that direct impact on spacc-craft will obtain a statistically rneaninglul s,unple. The leartexpensive remote sensors arc Eanh basedi coosequently, these

Dorutld J- Kessler

sensoni have provided lhe besl data to datc. The measuremenls!o datc havc been obtained using gmuad telescopes, groundradar, and returned sDacccraft surfaces.

Ground telescopes

A I cm diameter nr€tal sphere in sun light at 9{X) km distancewould appear as a l6th magnitude star. Since telescopes largerlhan about 30 inches can detect stars of this magnitude, in1983 NASA Johnson Space Cenrer (JSC) conr.acred MITLincolo Labs to use their Experimental Test Site (ETS) to lookfor I cm debris. An advantage of the ETS was that it conlainedtwo 3l inch telescopes which could look at the sane area ofthe sky to use paralliu to dete.mine altitude. lt was lblt thatthis feature would be essential to discriminate against theluminosity caused by much smaller meteoroids hitting theEaflh's atmosphere (meteo$) at about l(X) km al(itude. Thetelescop€s were op€rated just after sunriet and just before sun-rise, when fte debris wirs in sun light and the telescopes werein darkness. The telescopes were pointed vertically. and deb.iswas obs€rved to pass through the 6eld of view. Nine hours ofdata werc recorded on video tape and analyzed by LincolnLabs. Published resulls (Taft, et al., 1985) concluded lhar rheETS detection rate was 8 times the rdte expected from objectsin the catalogue. However, two errors were found in the analy-sis. plus one ofthe assumptions proved to be wror|g-

NASA, JSC reaoalyzed the ETS data and found parallax

errors wbich placed a larger numb€r of the objects detectedinto the category of m€teors. Thi$ reduccd thc delected orbital

debris to between 2 and 5 times $e catalogue rate. dependingatmosphenc se€ing conditions. A calibratioD enor placed thelimiting magnitude of the telescopes at 13.5 for debris wirh thetypical angular velocity of 0.5 deg/sec. Finalty, independentmeasuremenls u$ing radar. infrared wavelengths and opticalwavelengfis determined that the assumption that debris frag-nrents would rellecl light similar to a metal sphere was wrong.Debris fragments renect much less light than a metaisphere...typically only abour lO"Z of the lighr is reffecred.although some objecLs reflect a larger fraction. Consequently.the liniling size measured by lhese telescopes was about 8 rcl 0 cm.

Since then, NASA has worked closely with the US SpaceCommand to us€ their Groutd Electro-Optical Deep SpaceScnsor! (CEODSS). which are telescopes. similar ro LincolnLab's ETS. except they are slighrly less sensitive (limiting

magnitude of about l3 at 0.5 deg/s€c.), and have rwice rhe fieldof view. Over a hundred hours of data have been analyzed byNASA which produced nearly a thousand orbiting objccrs. TheUS Space Commaurd caralogue was used to predict which ofthe detected objects were already in the caralogue. Only abouthalf of thes€ objects can be identified as being caralogucdobjects (Henize, 1990). Consequently, these telescopic mea-surements have provided convincing data to NASA thar alaboul the l0 cm thrcshold. the low Eanh orbit cataloSue isonly atDut 50% complete. The exact limiting size measured by

Thc current and future environment

l0'1

1e4

MET€OROIDS

10 '7

10.2 10-1

t9s PnEOTTED O€8R|S

io3

to@

tt

10-s

t 10r 102Diameter [cml

Cumulative llux in 1995 b€twegn 600 and l tOO km altitude

triturc 3r Early model predictions of orbital debis cnvircnm€nr. (published in l gg l ,

each ot thes€ two differcnt types oi s€nsols is a function of twodifferent distribuiions relating signal rctum to size, and also afunction of the raic of increase in the number of smaller debriswith decreasing size. When these disfibutions are taken intoaccount. the "average" limiting size of thc t€lescopes can beshown to be slightly smaller lhan rhe "average" limiting sizefor the catalogue. The DOD has a progrlm to b€ner understandtle size and orbits of these uncatalogued objecls by trackingtienr with radar aod ground telescopes. However. trcauseNASA .equires data on smaller debris, its program has beenexpanded to obtain measurements using the Haystack groundradar. This radar is now detecting a smaller Frpulation al a ralethat is morc than 60 ttmes the catalogue rate.

Crourul radars

The failure ofg.ound telescopes io detect I cm orbiting debrisforced NASA ro re€xamine the use of radsr to detecl utrcata-logued objects. A major rcason that US Space Commandmdars do not detect smaller objects is thal mosl of their radarsoperate at a 70 cm wavelength; consequently. objects as smallas a few centimeteni in diamerer are well inro the Rayleighscallenng region afid rellcct a very small fraction of rhe mdarsignal. A shoner w:rvelen$h radar, in principle. colld d€tectsmaller object\. Ho\rever, another factor in the limiting size ofcataklgued objects is id lhe merhod of the operation of radarsto catalogue objects. In order to catalogue an objecl, the objecrmust finit be found within a large volume of space. then the

object must be tBcked by several radars over an extendedperiod of tirne. This process requires a larger rhreshold sizethan a process which only required detection within a smallervolume when th€ object is close$t to the scrf,sor.

tn 198?. NASA, JSC dcvelopcd a technique of using a rndarin a "beam pa*" mode, where $e rddar star€s in a fixed direc-tion (preferably venically) and debris randon y pass€s throughthe field-ol-view. In this mode, using a relatively incrpensive,high powered, moderate size X-band radar (3 cm wave-length), objects as small as I cm could be deaccted at I500 kmalritude. ln 19E7, interest in the haza.ds of orbital debris to fteSpace Station produced a series ofevents wlich resulted in anagreemenr betwe€n NASA.4SC and the US Space Commandto operate the Haystack radar, localed n€ar Boston. in rhebeam park mode. and to develop the nccessary computer pro-grarn$ to aoalyze the data. However, the Haystack radar is noroprimally designed (the antenna beam width is small. conse-quently more time is required to obtain the necessary data), no,optimally placed (its too far north, and cannot s€e low inclina,tion debris). Therefore, a Haystack auxiliary radar is beingbuill ncxt to the Haystack radar. In addition, another radar nearthe equalor is m be built, altiough the detail of this radar hasnot yet heen resolved. As a rcsult of this s€ries ofevents, a sig_nificant amount of grcund radar data has b€en obtained usingthe Arecibo, Coldstone. and Haystack radars.

To rest the concept of obtaining orbital debris data in a b€ampark mode, in 1989. NASA, Jer Propulsion Laborarory (JpL)

23

24

used the Arccibo Observaiory's high-powcr S-band ftdat, and

the Goldstone Deep Spac€ Communicaiions Complex X-band

radar to obtain orbital debris data. Neith€r .adar wa,s optimally

configured to obtain data in this mde, although borh radars

were Dredicted lo detcct small debris. if it existed. lo l8 houni

of operation, the Arpcibo experiment detected n€atly 100

objects larger than at estimaied 0.5 cm in diam€ter (Thompson,

et al., 1992). The predicted nudber from th€ catalogue alone

was akrut one- ln 48 hours of observation, the Goldstone radar

detected about 150 objects larger than about 0.2 cm in diarneter

(Goldslein. et al.. | 99)). The probability trai at least one cata-

logued obi:ct would pass thmugh fie n€ld of view during the

48 houn was about 0.13, indicating a population which is

slightly more than 1000 times the catalogued population.

Bocaus€ little effon was made to accuratcly define thes€ radats

lield of vie*, and to pmperly calibrare tlre radani, this data has

fairly largc uncenointies. Even so, tl|ese two experirnents did

demonstrate that data could be obtained is this mode of opera-

tion, and that there was a large population to be d€tected.

After testing the concept, NASA committed to a program of

usinS the Haystack radar to obtain orbital dcb.is data. The pro-

gram included calibration of fragmenl size using a iadat range

and fragnrcnts from ground tests, calibration of th€ anteona pat_

tem. developm€nt of a real-tirne Pr(rassing and Control

System to process and recend deiections. aod establishment ofa

dam processing facitity at JSC. In onder to ensurc that NASA

was properly acquiring and analyzing the data, a peer review

oanel r.vas established. The chairman was Dr. David K. Banon

aod included other well known expens ftom the iadar commu-

nity. The panel concluded that "lhe 6ital Debris Radar

Measu.ements Projecr is fundamentally sound and is based on

good science and engineering." They also made a number of

recommendations to improve effciency or accuracy of the data.

Many of those recomrDendations havc bcen impleinented.

To date. over 1000 hours of data have been collected, of

which over 800 hours has been analyzed, and more than 2ffi0

objecLs have be4n detected passinS through the radat beam

lStansbery. et al., 1992). Figure 4 gives tlte altitude of each

detection when the radiu beam is parked in its most senltitive

position of ltrking venically, compared io the detectiot rate

cxpe{:ted for the catalogue alone and the rate prcdicted by the

model given by Kessler et al., 1989. At the lowest altiludes

(350 km), objecls larger than 0.3 cm are detected. Al ihc high-

est latilude ( l4OO km), objects larger than 0.6 cm are detected.

The detection .dte averaged over all altitudes is aknrl 65 times

the rate predicted by the catalogue alone. In lhe altitude band

between 850 km and l0OO km, the rate is 100 tirnes the rate

predicted by the catalogue alone.

Re&)reaed sumples

Objects returned fiom spaca usually contain pits ot holes fmm

hypervelocity impacts with meteoroids or orbital debris.

Outside the laboratory, these are the only two pos.tible sources

which can impact surfaces with suflicient v€lo.iiy lo cause

Dourld J. Kessler

mcltilg of the surlace in the impaclcd ared. One techoique lo

determine which of $es€ sources caused thc impact pit or hole

is to use the sranning eleclron microscope (SEM) dispe.sivc

X-Ray analysis to deternrine the chemistry of material meltcd

into the surface. This analysis has becn completed tbr somc of

the pits found on Space Shuttle windows. impacts inlo surfaces

retumed from the Soliu Max repair mission in 198.1, surfaces

on the r€tumed Palapa sa(ellile. and some of the Long

l)uration Exposure Facility (LDEF) surfaccs. retumed t() Eanh

in 1991. Becaus€ LDEF was a controlled expetiment. was it

space for nearly 6 yeiuri. had a large surface 4sa. arrl was

always orient€d in the sarne dir€ction with respect to the

orbital vclocity vector. these surfaccs are providing thc besl

data to date. Analysis of LDEF surfacrs is still continuinS.

however the dala analyzed thus liu exceeds the quality of the

eadier drta.

The largest impact crate. predicted and found on LDEF was

slightly larger than 5 mm in diameter. likely due to an imPact

ty an objgct I mm in diameter The number of impact craters

increased rapidly with decreasinS siz-e, with more than 3(XX)

crateru larger than 0.5 mm. The most complete chemical

analysis has becn conducted by Frcd Hitrz, the hincipal

lnvestigator for the Chemistry of Microineleroids Experimcnt

(Hitrz. et al., | 991 ; Bemhatd. cl al.. 1992). The analysis to date

indicates that about l5olr ol the impacts in the gold surlaces.

facing in the rear direction. are otbital debris. Thc most com

mon orbital debris impacts are alumioum: however. copper.

stainless steel. Daint flecks. and silver were also tbund- Orbital

debris impacts on rear surfaccs was a surprising rcsult because

a very snrall set of elliptical orbital debris otbitt are capable of

hitting the rear surl'aces (Kes$ler, 1992). The most p.obablc

direction for orbital debris to impact is the fionl and side sur-

faces; the surfac-es lbcing in this direction are made of alu-

minum, and aluminum impacts cannot be idenlified. Even so,

14% of th€ impacts on these su.taces wete identilied as orbital

dettrisi the origin of 55% of the impacts could not be identified

because only aluminum was detected. lf the ratio of aluminum

to o$er orbital debris compositions found on the gold sudaces

is also on the aluminum sutfaces. then mosl of the impacls on

aluminum sudaces that could not be identitied would have

been caused by atl aluminum impact. This would increi..\e the

orbital debris impacts on the rluminum surfaces. When aver-

aged over all orientations. the averaSe numbet oforbital debris

cmters to meteoroid cmteni may be ak)ut the same. or slightly

higheq as the results obtarned from the Soliu Max Satellite

(Barrett. et al.. 1988). Analysis of the I,DEF surlaces is nol

completc.

PerfiaFi the most surprising result on LDEF came from the

Interplanetary Dust Experimcnt (Mulholland, el al.. l99l).

This wa^s the only exp€riment on LDEF which measured the

time of impact. Six detectors were on onhogonal surfaces. and

sensitive lo impacts smaller than I micron. The surprising

rcsult was that mos( impacts could be ass(riaied with "orbital

debris swarms.' That is. the sensors would detect a larSe

Tle currenr arul future cnironment

t0 .0

- HAYSTACX OATA- . Pfi€DCIED IU X){71

_ c[cuurEo cATAtoGEo FAT€

25

t . 0o

E

o

o

oo

o

o

0 . t

0 .01

0.m,50t) t0m

Altitudo lkml

Ii8ure 4: Ratc ol detection b), lhe llaystack Radar, venlcally F)int€d for lSll houni.

increase in llux. lasting lb, a few minutes, a{ the same points in

Oe LDEF ort'il. and thes€ points would slowly change with

timc...chalacteristics of orbital precession rntes. However, in

rctlosped, these results should not have been such a surprise.

The small amount of mass required to produce a l?trgc fluxof less rhar I micron detrris in F-:nh orbit. coupled with their

shon orbital lifetime would predict that a large number of par-

ticles could be found in orbils ck)se to the orbit of rheir sourc.e.

Ifthe source is paint being removed by atomic oxygen erosion,

then less than I0 grdms of paint is needed lo be removed from

each orbiling spacecraft pcr ycar kr explain these results. lfthe

source orbil is highly elliptical, then less than I gram of paint

need is needed. These are rates consistent with the ratesetpected liom atomic oxygen erosion-

Other sourccs are possible. such as the large anKlunl of alu-minum oxide dusl that cach solid rocket motor exDels whenfired. This dust is expelled at a velocity of3.5 krn/sec. and overa range of directions, most of which would cal|.se the dust toimmediately reenter the Eanh's atmospherc. Altbough some

dusl would remaiD in orbir. mosr should reenter quickly aidnot producl swams lasting for seveml rnonths. as observed.

However. Lhis is n()t to tay ihal a spenl rocket stage might notslowly releasc suflicienl dust lo produce lhe hng lastingswarms.'l hese are Xrssible arear of future research.

Summar,- environnent

A sunrmary oI the besl measurements l() date is shown in

FiSure 5. compared to the natural meteoroid environment.

When comparcd to the 1995 modcl predictions shown ioFigurc l, there is a general agreement (within a factor of 2)over nearly the entire size range, even though some of theme:$uremedts were made at altitudes of 600 km. or lower.where d€bris was predicted by some analysis to be muchless than the environment between 6(X) and ll00 km_ Thiscould be interpreGd as an indication that the cnvironment wa.sunder predicted for sizes smaller than about 5 mm, and thismay be the cas€. However, sone of this data is irdicating thatelliptical orbits are imponant in this size range and at theselower allitudes. Elliptical orbits will produce an environmentthat is less altitude dependent than circular orbits. Until allof these parameters are understood, or until small debrismeasurements are made at higher altitudes, there will be anuncenainty in extrapolating thes€ measurements to highe.altitudes-

What should have been the easiest Drediction is the cata,logued popularion. The prcdiction was lhat would be slighrlyover I l(m catalogiued obj€cts ir orbit by 1995. The.e are cur-.ently 7000. and this number might increase ro ttfi)o by 1995.but not | 1000. Therc are three important rcasons for undcr pre-

dicting the catalogued populatioo: l. RarlEr rhan a slighlincrease in the amount ofmatgrial launched inlo sDace. the raleof launches world wide has remained constant. 2. Two of theth.ee highest solar acrivities in 2m yeafs of record keeping4-cured bctween 1978 and !992. High solar activity increasesthe atmospheric density. causing more objects to reenter fromorbit. 3. Since 1981, the US has lead an effo.t to minimize on-

/ Dobfh Swsms m IDE arFrinent (odrepoldod horn smaloa inpds)

LOEF, 1984 - 1991330 lm - 475 km

Orbld d€bris lxrd o.t }f,tr: uoariT5rf

metsofoids o 500 krn(Griin, er el., 1$51 /

26

106

t05

!o'

to!

lo2

io l

100

t o l

S

(E

E

Ec

oi,

Ioo.:-g

E

o

Donald J. Kessler

breakups. The major issue is the size and velocity distributionoffragmenls produced as a function of the amount of fragmen-tation energy by various fiagrnentation energy sources. In thepast arnd near term, the energy sources have mostly been chem-ical, and we can see that this sowce has already produced ahazardous envircnment for most spacecrafl. Chemical explo-sions can easily be contsolledi so our near tcrm environm€ntwill b€ a function of our efforts toward eliminating lhes€ paslsources of explosions in orbit. How€ver, in the fu(ure. riemajor energy soufce could be kinetic energy; this source is notas easily controlled. The amount of kinetic en€rgy representedby an object as small as I kg, traveling lO kr/sec is nol roodifferent than the amount of chemical energy which causedpa$ chemical explosions in orbir.

Lile most chemical explosions. most of the muuss of frag-ments from a hypervelocity collision is in the larger fragments;however. because the eneagy source ts concentratect rn asmaller amount of mass, higher temperatures are rcached andmelting of the impacted spacecraft occurs, which resuhs in asmall, but significant fraction of the mass being distribured insmaller fragmcnts. These characterisiics of hyp€rvelociry col-lisions, combined wirh the irrcreasiog rate thar rhey could(rccur if no changes in curcnt practices are made within the

#id' to '

cliarnetor lcml

Flgurc 5: Meteomid environmenl comparcd to recent m€asurements of orbital debris envtronmcnr.

orbit explosion. As a result, fewer US, ESA and Japaneseupper stages have exploded in orbit than would have occunedifthis effon had not been undenaken.

For the abov€ three reasons. the 1995 prediction of theuncatalogued population should have also been high. That is,as a result of fewer large objects in orbit. there should be fewerrandom collisions to generale small debris; also as a result offewer on-orbit explosions. there should have been less smalldeb.is. The implications are rhat either the current breakupmodels are under prcdicting the amount of small d€bris whichis generated and remains in orbit, or there are unmodeledsources ofdebris- The most likely caus€ is the brcikup models,&rd that these models are und€r predicdng the fraction of masswhich Socs into smaller debris. This should not be surprisingsince smaller fmgnrents are Inore likely to be tost during

Smund expenm€ns to determine breakup characteristicli, Thepossible tendcncy of current breakup models to under predictshould be keep in mind a.s similar models are asain used topredict the fulure cnvironment_

Futurc environmena

The most important parameters in predicting tlle future orbitaldebris environmeDt is the rate and cons€quences of satellite

The current andfuture environmenl

next fcw years, ma&e lhem imponant to the futurc ort ital&bris environment in low Earth orbit.

Thc Defcnse Nuclear Agency (DNA) has a program rodefine bener breaftup models (Tednerhi, er al., l99l). Theanalysis of hypervelocity tests by DNA ate not yet complete;however. a gross characterization would be thal they are ingeneral agreemenl with the assumptions of€aJlie, models; thatis. at l0 km/sec. a I kg fiaSment can complerely fragment al(Xn kg spacecraft, producing hundreds of I kg ftagments,cach of which could fragment another spac€craft, and also pm,ducing millions of smaller hazardous fragments, each capableof damaging an operational spacecraft. Some masses cata_logued by rhe US Space Command are smaller rhan I kg,while there is undoubtedly some uncatalogued fragmentswhich exceed I kg- The rate rhat catalogued otrjects can b€expecled to collide wilh one another is a fairly easiiy calcula-Iion. Based on the current population, th€ .ate is about one col-lision every 20 ycars (Kessler, 19914); however. to date. suchan event has not been observed to occur. Estimates ofth€ massof uncatalogued debris predict that the current rate of satellirebreakups from hypervelocity collisions is about once every g

years (Kessicr. l99lB): there is data and analysis (Johnson. etal. , l99l: Johnson, 1992: McKnight. t99l) ro suggest thatsuch cvents have nccurred. Becausd these rates are propor-tional to the square of lhe number density of objects in orbit,these collisional fragmentation rates can trcome much morefrequcnt in the relatively near future if objecls conlinue toaccumulate at past rates.

The Earth's atmosphere will remove fragments from lowEanh orbit. If fragments are removed faster than ihey are gen,erared. rhen an equilibrium environment will be esrablished.and this environment will not increase unless new material isadded to Eanh orbit. However, if collisions are producingfragments at rare faster than they can be removed by theatmospherc. then thc orbital debris environment will conrinueto increase even without adding new material to orbit. Thesatelhle population density which will pnduce collision fragmenls at the same rate they 4re removed has been defined as a''cntical density. To i![empt lo maintain a satellite populationabove the critical den$ity means that debris will increase as aresult of random collisions alone. Objects nearer the Eanh areremoved al a lasler rate. The aate that fragments are generatedis not only a function of number density, but satellite size andlncl ination of the orbits.

Figure 6 compares the 1989 catalogue with a calcularcdcntical density as a funcrion of ahi ludes (Kessler, I99l). Thefigure shows thar below EU) km, ahospberic drag removesfragments at a sltfticienlly large rare and rhal this region is welltrelow the crilical density line. Between 8U) km and l0OO km.the currenl population density is atnve lhe critical densily linc.Atbve I (|(X' km rhe physical sizc of satellires is smaller. rcduc,ing the lragmcnt genera(ion rdle: even so lnrs rate exceedstlre removal rate again atnve l40O km. The uncenainly inthis cntical density line is about a t'actor of 3i consequently.

within this uncenainty is the possibility thar a crirical densityhas not yet been reached. However, even if it has not beenreached, the population of these two altitude bands might beexpected to exceed thc critical density within the next fewyea6.

Similar conclusions have been independently reached byotier reselLrchers using different modeling approaches(Eichler, er al. , 199O, Talert, l99l:Farinel la. et al. , l99l I tee.et al., l9q0). An evolurionafy model (EVOL) develop€d byNASA, JSC (Reynolds, er al.. 1990) illustrates the same rrendsard is being us€d to evaluate the consequences of various pos-sible operational practices. In Figure 7, this model is used roplot th€ I cm population at 400 km and l0(n km as a functionof time for several operdtional conditions. The ,.case l" curvesassume "business

as usual, keeping the world launch rale at rhecurrent lm launch€s [rer year, with all objects allowed to accu-mulate in orbi{ at the end of their life, and no reducrions in rherate objecrs explode in orbit. At rhe end of 100 yea$, rhe envi-ronmeni i! a factor of l0 larger than thc currenr envtronmcnt allom km altirude. At 4OO km alrirude, the nux varics due ro thevar-ving atmospheric density due to solar activity: even so, theflux has a general upward trend with time. The ,.case 2', curvesassume the same conditions, except all chentical explosionsare eliminated in the year 2000. Note that the rate of increaseof I cm debris immediately begins to decreasc when explo_slons are eliminated, especially ai 400 km. However, by theyear 2030, rher€ are sufficienr old rocker bodies and payloadsin orbit that the satellile break up rate tron random collisioncauses the I cm flux to begin to increase again. By the cnd oflhe century at l(no km the fact that explosions were elimi_nated is of minor imponance. Under these conditions, kineticenergy has become the most imponant source of energy caus_ing salellites to break up in orbit.

Kinetic energy can be eliminated by eliminating mass inorbil. The bottom curve assurEs the same conditions as the2nd curve. except that after the yea. 2000. rocket bodies arerequired to reenter after delivering their payload, and afte. fteyear 2030, payloads rtre required to be removed from orbit atthe end of fteir operational life (trssumed lo be lO ycars).Under thes€ conditions. the I cm environment continues todecline. Consequently, this model, as well as orher rnodels. ispredicting thar furure payloads and rockel bodies must be reen,tered in the relatively near future in order to keep the futurermall debris population from rncreasing.

Environment mrtr.sgemenl of lotd Eartb orbiaIn the shon tern, managemenl of the low Eanh orbit orbitaldebris environment is concemed with the control ofexplosionsin orbit. Users of space are already adopting new operationalprocedures which are designed to accomplish this. Thisincludes depleiinS rhe unused fuel from orbiting upper stagesafter payload orbit insenion and conduclinS military tests atlow altitudes to ensure that all fragments reenter. However. itnow seems clear (hal in the long term, these changes will make

28 Donul,l J. Kes:'ler

! to-a6

E

o

6 S

t F

tsin d

EF

i xi =

Adlu.Ld C.ilicalDon8lly

. Assumes Inclinatbn and szedi9lributlonE at varioos altitud€sdo€s nol charEs with lime.

. At a lovgl abova the c,itical dsnsilvlino, populelion growtn lrom raMorncotlstons can bg €xDecled.

linle difference and new, more costly changes must be made.

Some of these changes should begio now.

The design process should begin now which would prevent

mass from continuing 1o accumulate in cenain regions of low

Eanh orbit. Various techniques hav€ been studied vhich could

accomplish this (I-oftus, et al., l99l). Thes€ techniques

include the planned r€entry of rocket stages, payloads, selfdis-posal options, retrieval, and the us€ of drag devices. An option

Dgc '89 C8irlog

which could bcgin very soon is the plalned reentry of cenain

rocket $ages. With a minor amount of additional fuel and an

extended battery life, many of the currently uscd upper stages

could be left in orbits with much shoner lifetim€s alierdeliver-

ing their payloads.

Upper stages lefi in t ansfbr orbits from low Eanh orbit to

geosynchronous odlit can cause a signiliciurt problem to ahi-

tudes belo* about 500 km (Kessler. 1989). This is because at

10-e1500 2000

Allituda (km)

l'lgure 6: Critical dcnsity compared to 1989 calaloged population where the cahk)g is maintained.

l c a

l C a

1 0 - 3

r o ' t

r f t

VEAR

FlSuar 7: Predicte{t I cm lhx for three possible operational practices, allassuming a continuation ofcuFent launch rates.

1000

EI/OLVE MOO€L PRo'ECTPNS

CE€ |. &.6ics a.C6.2- E{ry daDrB ntgi-n: dir|rle

arpbsrr3 ra. yd 2qnC|..s aoCtetlr d.b.b nr||!An(r'

dirh.t spoLsr! r|b. F 2(!oslirad. $F. *aC! Etrudion dt . t€t 2OO

Tlv currenl und future enyironmenl

thcse lower altitudes, the orbital lifetime of thes€ high energy

orbils can bc vcry krng. Howevcr. the lifetime of these stages

can b€ shonened considerably by a very small change invelocity at the aFxlgee of the transfer orbit. The required

change in relcrity iri so small lh"t lunar and solar gravity canbe used. if the original orbit has the proper orientation with

rcspect lo the sun (Mueller, 1985). to cause the uppe. stage toreenter within a ycar.

Thc reentry of paykrads is more diflicult re*ause paykrads

do not usually have the potential propulsive capabiliry of

upper stages. Altematives include the use drag devices (i.e.,

deploy a large surface area. such as a ball<nn, so that lhe

atmosphere will d.ag the objecl out of orbit rnore quickly), orlelhels. The cosl of ahemativcs such as these and others need

to b€ evaluated lgainst thc cost of simply increasing payload

propulsion capability.

The cosi ofthe retrieval ofobjects in ortrit using current tech-

nokrgy is very high. and this is a nrajor reason lbr designing

renxrval options into futurc nL-kets and payloads. However,

this is nor lo say iha{ fiture technology could not provide a ded-

icaled relrieval system lhat would be inexpensive to olrrate. Ifyr. such a system might thcn rclieve sonrc future design

requiremenlr. However. the need to nol lcare dead rockel

stagcr and payloads in low EaJlh orbit is immcdiate, and should

becomc u dcsign consideration lbr all new programs.

PART II - GEOSYNCHRONOUS ORATT

Orbilal debris sludies conceming geosynchrunous orbit urc

sl ightly r )re than I0 yeani old (Hechler, et al. l91l{)). Mosl of

the studles () date have been concemed wiih the larSer, crta-

logcd objccls (ChoboIov. 1990; Flury. l99l ). No models havc

been developed lo predict lhe populalion of small dchrir, nor

hon this B)pulalion might vilry with timc. No mcasurcmcnls

ha!e been conducled to dctermine the orbital dcbris population

to sizes snrallcr than akrut I nreter in diamcter.

The reasons lbr thc lack of modeling and data dre t\roibld:

L The higher col l is ion velG-it ies in low Earth orbrt cause Ihe

conscqucnccs of collisions t() be more dramalic than in 8eo-synchronous orbil. For some time there has bccn sullicicnt

dala ttr show that the environment in low Eanh orbi l al l icted

the dcsign of planned NASA missions. result ing in more

resources being devoted (r undentarrding and controlling lhis

environmenl- 2. Geosynchronous orbit is fanher away tiom

funh. Whilc this distancc har kept lraflic to geosynchronous

orbit smallcr than to low Flanh ()rbit. it has also made obrrva-

t ional data more dif f icul l to obtdin: consequenlly, morc

resorrrces may be required to ohtain the neccssary data than

havc heen de\'otcd to bw Ennh orbrt rcsearch. Dcspilc the

lack of dala. sonle uscrs are unilulerally adopting a policy of

moving their dead payloads to higher altitudcs. "grave-yard"

orbits. While this may provide some operational conlenience.

it mil!, be t()tally inappropriule to long-term envir()nmcnl

managenlent-

29

Fundrm€narls of orbltsl chmcterlstics lo g€odynchronousorbltThe orbital period of an obje.l in gaxynchronous orbit iri thesame as the time rcquircd for the Earth to spin one revolulion.or about 23 hours and 56 minutes. lf th€ orbit has zeru eccen-tricity and inclination, the object will appear to be stationary(hence the orbit is sonetimes referred to ar geostationaryorbil) over a location on th€ Ea(h's equalor, and b€cause ilsaltitude is about 35.786 km, it can be seen by nearly an entirehcmisphere. Howeve., this orbit is suflicieDtly fiu from theEanh that the forccs produced by the sun and moon will notalbw the orbit to naturally maintain a zero eccentricity andinclination. Eccentricity is typically less than O.(nl, keepingpengee minus atpgee distaltces to less thar l(X) km. This issufficiently small that il does nol cause a ground tracking prcb,

lem. However, inclination 1,ill nalurally oscillate belween zeroand l5 degrees over a 53 year cycle. This can caus€ a gruund

lracking problem, so that a significant fuel budget is usuallyrequired for "North,South stalion keeping."

Without "East-West stalion keeping," another ground track-

ing problem would exist. The fact that the Earth is not a pertbct

sphere also causes an (xcillation in thc longitude of an uncon-

trollcd satellitc. This oscillation is abo{t the nearest "stable

point. localed over longitudes of 75 degrees East, and 105

degrees West. lfthc dcsir€d position of the salellite is far away

from the stable points, the satellite would oscillate nearly

halfway around the Earth before rcturning atnut 3 years later.

l.bnunarely, East-Wesl station keeping fuel requirements arc

small. requiring less than 5% of tie fuel budget required for

Nodh-South station keeping (Flury, l99l).

The orbital debris problem

Thc orbital debris problem results from the accumulation ol

satellites, fragnrents of satellites, and operalional dcbris in

orbits which pass through the paths ofoperational geosynchnr-

nous satellites- Any of these objects could collide with an

opcrational spacecraft, damaging it and reducing its opera-

tional litc. lf a large numher of objects accumulate. the hazardcould significantly add to the hazard! from other sources. such

as collisions with narural meteoroids. Figure 8 is an expansionof FiSure 2 for 1991, and comparcs fte flux of catalogued

objects in low Eanh orbit wilh ge{r..iynchronous orb(.

At georynchronous altitudes, theae is only one naluralp.(ress which will eventually eliminate a satellite from thisaltitudc. C)ver exterded periods of time. spacecraft and frag,

ments ol- spacecraft will break up tmm collisions with orherobjccts which arc either in or pass through the geosynchronous

reglon. The smalles( fragments (less than about IO micronsl

are atl-ected by solar radiation, which both incrqrses thc orbiral

eccentricity and dccrca.ses the orbital semi-major axis. result-

ing in the smallest fiagments being removed from orbit by hir,ling the Eaih's atmospherc within a few months (Mueller. etal.. l9ll5: Friesen, et al.. 1992-8). Ifthis p.ocess acted quickly

to remove fragments of nll siz-es. th€n the accumulation of

The current and Juture environmerTl

lirst bking the ratio of the c'ollisional cross-s€ctional arcaof active satellircs to the arca of tlrc geosynckonous ring.

lnactive sarcllites, which would generally have an inclination

of several dcgrees, would pass thtough the geosynchronous

ring twice a day. The collision rate per 12 hours is then this

ratio times the number of inactive satellites. A morc accurate,

but morc complcx approach was developrcd by Kessler(Ke$rler. l98l). Both approaches assume ar| even distributionofsatellites around the geosyncluonous ring. Perek's appmach

caD be shown to be equivalent to Kessler's approach as long asorbital inclinations of uncontrolled dehris are between about

0.1 degrees and 20 degrces, and orbital ccceltricities are small,

which is the cas€ fo. most debris in geosynchronous o.bit.

The simplicity of Perck's approach provides sorn€ addi-

tional insight; namely. thal collision probability in geosyn-

chronous orbit is approximately independent of orbital inclina-

tion. Conscquently, this approach can be us€d to obtain the

collision rate betwe€n any two inaclive satellite$ if the differ-

ences in inclination arc between 0. I and 20 degr€es, which is

generally rue for most pairs of uncontrolled satelliies. If N is

the number of uncontrolled satellites, then the average rnte of

collisions between any pair would be N(N-l)/? times the rate

ofcollisions between a single pair.

There are currently about 250 objecls known to be in geo-

synclronous orbit, with average linear dimensions of about

5 mete$ (Royal Aircraft Establishment). Most of these objects

arc in orbits which confine thcir motion to a 100 km band.

cenle.ed at the geosynchronous altitude. Assuming all of these

objects become uncontrolled and are randomly distributed

within this band gives an average collision rate of once every

l5.tn0 years. This is about the same rate if one a.rsumed an

avenge llux of 7 x l0e/mr - yr. From Figurc 8, this flux

might be an appropriate average within ahe g€osynchronous

band. although it is likely tha{ the appropriate avemge could

be slightly higher due te the non-uniform distribution within

the geosynchronous altirude band- Even so. the collision.

rate is low compared to the rale in low Eanh orbitl however,

like low Eanh orbit, it is high compared to the rare that objects

are removed by natural forces. Also like low Earth orbit.

this rate incrcases as the square of the number of objects in

orbit: consequently, if the mte of accumulation of objects in

geosynchronous orbit continues at its curent level of 25

objecls per ycar. then there is a 50% probability rhat there will

be at least one collrsion in geosynchronous orbit in th€ next

140 yea.s.

These rates are smaller than olhcr published ratcs... in some

cases. signilicantly smaller. Pa( of the reason is in lhe assump-

tion that the satellites are nmdomly distributed wi$in the geo-

slnchronous akitude band...they,ue not. Some researchers

have obtained significantly higher collision .ales at cenain

longitudes (Gucrmonprez. l99O). However. these higher rates

may not be representative oi the Seneric hazard. but result

liom rhe desire lo maintain the satellite over the same lonSi-

rude. These higher rates are reduced signilicanlly simply by

J I

terminating siation ke€ping. Once station keeping is termi-nated. the satellire begins to drift in longitude. and the distrib!-tion of satcllites approaches a more uniform distribution.Researchers who assume that the sarellites are simply aban-doned (Hechler. 1985) obtain collision rarcs that are less than afactor of two differcnt than the collision rales obtained by

assuming a uniform distribution. Consequently, the long-termcrror in assuming a uniforrn disu'ibution is probably small,although this assumption should be carefully exanined.

when the size and number of satellites in geosynchronous

orbit is assumed to be largc, the collision rate will a.lso belarge. For example. a rate of one collision every 4{X) years to600 yeani. and a O.16 probability of a collision over a 20 yearperiod was calculated by Hechler (1985). and is frequenrlyquolcd by others (ESA, 1988; Flury. l99l ). This apFears ro bevery different than the one collision every l5.0OO years previ-

ously calculated. The primary reason for thes€ large differ,ences is in the very largc size and larger number of objects ingeosynchronous orbit assomed by other aothors. These authors

somelimes assumed 2m salellites with linear dimensions of 50meters, and &s many as 10,(n0 one cm orbiting fragments ingeosynchronous orbit. Existing satellites in geosynchronous

o.bit are much smaller, and therc is no hard data describing thenumber of small fragments in geosynchronous orbit. Even so.

a comparison with the natunl hazard reduces the ambiguity

introduced with these assumptions.

Collision ral,es are proportional to area for both m€teoroids

and debris; cons€qu€ntly the relative collision mtes from mete-

oroids and debris are the same for any assumed satellite size.

The collision rale from meteoroids increases with decreasing

neteoroid siz€; however, the darnage resulting from a collisiondecreas€s with decreasing size. Consequently, a high collision

rate with small debris may not be signiticalt to the over-all

hazard of lh€ spacecraft when compared to the rn€teoroid haz-

ard. A key parameter in comparing the debris hazard to the

meteoroid hazard is the collisronal velrrity.

Collislon velocity: meieoroids 8nd ssaellit€ brcrkup rsaes

lf all slalion keeping in geosynchronous orbit was terminated

so that orbilal inclinations could reach their naiural long-te.mdistribution. then the collision velocity that an average satellire

would experience would range for zero to aboul 0.8 kny'sec.

and have an irverage of about 0.5 kn/s9c. The average tnete,oroid velocily is l6 knr/\ec. Therefore. for a gtven mass. melc-

oroids will collide with geosynchronous satellires at about 32limes more momentum and | (Xn times more kinetic energy

than a collision with another object in geosynchronous orbir.

At 16 krn/sec, a 0.7 kg a meteoroid would trreakup the average

2m0 kg spacecraft. The rate that a 0.7 kg rneteoroid can be

expected Io collide with any one of the 250 satelliles, each 5meters in diameter. is about once every lo0,(m years (Zook.

1992). Consequendy, the cuFenl rate of satellite collisional

breakups is probably controlled by the current number of satel-liles in geosynchronous orbit. rather lhan meleoroids: how-

ever. Lhe time scale is very long betbre a satellite will breat updue to either ryp€ ofcollision.

Satellites in g@synch,onous orbit may break up morc fre_quently for orher reasons. In low Eanh orbit, nearly half of thecatalogued population is fragments ofsatellites, resuhing fmmmore than 100 explosions in low Eanh orbil. Most of rheseexplosions were due to the failure of an energy storage device,such as the tanks of upper stage which contained r€sidual fuel,or batteries on a spacecraft. Thes€ same potential sources areequally common in geosynchmnous orbil. Ar the rate lhatexplosions have (rccunql in low E:mh orbit. one should experctabout l0 explosions to have occurrd in geosynchronous orbit(Kessler. 1989-B)..,yet, non€ have been oflicially recorded(Johnson. et al-. l99l). However, rhere havc been two reponsof an obs€rver witnessing atl object exptoding in geosynchro-nous orbit- One rcpon wa-s from Russia. made in February,1992, reporting that in JuDe, 1978, a USSR Ekran sarellire wasphotographed as it exploded from what was believed to be aNickel-Hydrogcn battery failurc (Johnson, lE2). The otherwas on Feb. 21, 1992, when a Tital upper stage, launched onSept. 26. 1968, was video raped just after ir appeared toexplode (Bruck, 1992). However, as yet, no fmgments haveb€en calalogued from either of these events, which may not besurprising since fragmenls smaller than about I metet in diam-eter are diflicult to detect f.om the ground with sufficient .egu-larity to catalogue. Given the improbability lhat such eventswould be recorded, other, unrecorded explosions are likely lohave occurred. Cons€quently, a satellite breakup mte due tocurent ope.ational prilctices is likely to range belween onccevery I to l0 years...a rate much higlEr than the highest pre-dicted rate based on collisions. The nnal step in evaluaiing thesignilicance of thes€ breakup rates is in understanding thenumber, size, and velocity of fragments generated ils a resultof breakups and horr the resulting debris hazard compares tothe natural envircninent.

Co$equeDc€s of brcrkup in gecyncbroDoos orbitA brcakup in geosynchronous orbit has 2 possible conse-quences: ,. A brealup paoduces fragments large enolgh tobreak up atro$er intacl satellite. These fragm€nis contribute tocollisional csscading, or "a chain reaction" if an average ofmore than I large fragment per satellite breakup is generatedwhich stays in the gcosynchronous ring. lf the number of largefragments is significantly larger than ooe, the contribution tocollisional ca-scading will b€ greater. 2. A breartup producessmall fragr|ents that can collide with and damage operationalspac'ecmft. A key quqstion becom€s how this hazard comDareslo lhe nafural hazard.

Although some data is available on the numb€r. siz€ andvelocity of fragments generated as a .esult of breakups, mostof that data was genei'ated undcr conditions very different thanneeded to understand the conseqleflces of brcakups in geosyn-chronous orbit. Missing is data ftom collisions at about 0.5kn/soc. and completc data on explosion fragrnents smaller

Doru d J. Kes.sler

than l0 cm. Becaus€ satellite construc{rur ts more rmponanl atOe lower collision velocities expected in geosynchronousorbit. ;rny ertrag:lation of tests results leads to larqe uncenain_tics in predictions.

A "worst case" enviaonment can tle predicted by il\sumingthat only the ratio of target mars to pojectile mass. as deter-mined by hypervelocily lesrs, is imFnnant in prcdicring rheprojectile mass causing caaasfophic breakup at 0.5 kn/sec. Inthis case, a 5 kg p.ojectile could cataslrophically bi€ak up rhe2(XD kg satellite. C.ound explosions and explosions in spacesuggest that th€ largesl fraction (about hal0 of the salelliremass goes into about this size fragmenl (Kcssler. l99l-B). solhat about 200 fragrnents ofthis size would likely be produced.The same data also suggest that these fragmcnls would bcejected in all directions with an average velocity of about50 meteni/sec relative to the center of mass. This velociry issuflicient to spread rhe fmgments over thousands of kik)metcrsof altitude, so that al any on€ time. only about 20 of the 2U)fragments would be found within the lU) km altilude handwhere geosynchaonous liatellites arc located. Consequently,with this exfeme &ssumption. collisional cascading in geosyn-chro[ous orbit is possible; but with only 20 fragments persaGllite breakup to comnbut€ to the cascading and with thelirst collision not expected for l,l0 years, the cascading wouldbe v€ry slow. requiring thousalds of yearsi to be noticeable.

An equally imponant conclusion from this extreme assumltlion concems the "sale" dislance to place inactive satellitesoutside ofthe geosynchronous orbil. A satellite breakup whichoccuned wilhin a few thousand kilometeni above or below thegeosynchronous altitude would eject 5 kg fragments into orhitswhich passed thrcugh geosynchronous orbit. lf the brcakupwerc within a few hundred kilometen. the contribution to $ehazard to geosynchronous orbit would almrxt be as great as ifthe breakup had occurred within geosynckonous orbil(Friesen. et al.. 1992-8). Consequently, if the elergy of furureb.e.kup6 in geosynchrcnous orbit is not too different than Dastbreakups in low E.anh orbit. the safe distance to place inactivesatellites must be measurei in thousands of kilomeleni fromgeosynch,onuus ([bit in urdcr lo be el lect ive.

Hrzrrd resultiDg from explosionsSince collisions are not likely to be a signilicant source ofdebris in the near future, a more imp()rtant issue might well belhe consequence of past explosions in. or near. geosynchro-nous o.bit. Two ob.iects hale been obs€rved to explode: it isnot unrea.sonable that l0 times this number, o. 20 explosionshave occurred. Assuming lhe sam€ size aDd ejection vel(rityrelationships as b€forc. we should expect an average of 400additional objects. with mass€s of 5 kg or larger, to be in thegeosynchronous altitude band at any one time. If these lrag-ments are capable of catastrophic breakup of any of the 250geosynchronous satellites krown to be in gcosynchmnous

orbit. they would incrcas€ the catastrophic collision rate fromonce every 15.000 years to once every 8,300 years...i.e,, the

f!i

The current ond futurt' environment

explosion fiagments would be as importnnt as thc known satel-

lites in geosynchronous ortit in contrihuting toward collisional

cascading.

Explosions wrll also produce smaller debn$ which will causc

a haz.ard to other spacecraft. However. the type of explosions

which ue likely to havc (xcurred are nol likely lo produce a

large number ol small dcbris. [;or exaunple, a low intcnsityexplosion is predicred to produce about l00O fragnrents larger

ihan I gm ( Kessler, 199 | -B ). Thesr fragmcnt\ are iikely ro havevelocities largcr than the 50 nreter/sec for the 5 kg frdgments,

consequently, spread over a larger volurne of spac€...however.

dala sources giving the exp€.ted velocity is lacking. A conserv-

ative assumption would be that the velocities are the $ame,implying that about 100 ol rhe l(m fragrnents would be in rh€geosynchronous altitude band ar iuty one time. A total of 20

explosi{)ns would mean that 2000 fngments of I gm and larger

are in the geosynchronous altitude band al any one tirne,ptoducing a llux ol I impact every 18 million yeanr per

squar meters of spacecmft crosr o*tional area. The nr€teomid

nux for this mass is I impact every I million yea per square

meteni ol-spacecmfl cross sectional area. which is nxne than 5

times largcr lhaD thc dcbris flux rcsulting from these explosions.

In addition, given the low vel*ity of0.5 km/sec which debris is

likcly to collide with spacecraft in geosynchmnous orbit. a

debris mass between 5 and 25 tirn€s the meteoroid ma,ss

lchristians{]n. 1992) depending on spacecraft consruction, is

rqluired in order lo do the same damage to (he spacocrdft a\ a

meteoroid. Consequently, thc meteorcid Rux which is likely to

do th€ same damage 4r a I gn debris tr-agment is betwe€n I

impact per square meter every l20,fix) to 600,(X)O yea.ni, or

much higher than the possible debris nux rcsulting ftom 20 past

exDklst()ns.

All currcnl salellite breakup modcls predict lhat the traction

of satcllite nrass which goes into smaller sizes decreases with

decreasing size. (Jn the other hand. the amount of meteoroid

mass increases wrth dccreasing size. lf the debris flux of I gm

fragments is less than rhe mel€oroid flux, then all satellite

breakup mulels would predict lhal the meteoroid dux is also

iarger than the debris ,lux for sizes smaller than I gm- For theorbital debris hazard in Seosynchronous orbit to exceed the

meleoroid hazard. maoy limcs more than the ,lssum€d 20 satel

lites must hreakup in geosynchronous orbit.

Thcrelbrc. the ability of brealups ro prqluce an environ-

menl in geosynchronous orbit which is more hazardous thanlhe meteoroid environment is much less than in low Earthorbit. This should nor be inlerprered lhat (,ne should not b€

conccmed. b!l rather thal lhere is time to properly consider lhc

total environmental management issue in geosynchronous

orbit. and lo address the nrajor sourccs nf dchris in geosyn-

chroxrus orbit-

lnviroDanent management of gs)synchronous orbit

Thc only scriously considered lechnique t{) manage orbital

debris in geosynchronous orbit has been lhe use of a 'grave

yard" orbir (Sudderh, 1985; Choborov, | 99O; Flury. t99l).M(xit studies $how that ifan intact satcllite is placed in a circu-lar orbit about 200 to 3m km away from geosynchronous

orbit, it will suy therr. However, if one were to do norhing butnove all objects fiom geosynchronous orbit into such a grave-yard orbi(. th€ same orbital debris sources of explosions andcollisions would be taking place in the grave-ynrd orbit. Asdeveloped earlier, with only a 2U) to 3m km sepat"ation dis-lance, the orbits of fragments generated in the grave-yard ortritwould still causc an increa,s€ in the hazard in geosynchronousorbits that would be .educed by less than a facior of two com-pared to the h^zard caused try the objects fragmenting in geo-synchronous orbit. Several thousand kilometeni o[ dislance isrequircd in order to prcveot a signilicant fraction of satellitefragments fmm passing through geosynchronous orbit(Friesen, et al.. 1992-B). Consequently, it is important thal anylong term environment management include other elements.

Perhaps the rrcst imponant element is to minimize the pos-

sibility of accidental explosions in, or near. geosynchronous

ahitude. ln low E€rth orbir, this has been accomplished tbrupper stages by eliminating excess fuel after the upper stagc

has delivered its payload. Other energy slorage devices, suchas high pressure containers and batteries should also depletetheir energy source. Thes€ actions are marny onden of magni-tude more eflective at eliminaling ncar-term sources of debris

than is the usc ofa grsve-yard orbit.

However, in the lorlg term, the major energy source forsatellite fra€mentation is kinctic energy. This energy sourcecan only be elimimted by either eliminating the satellite mass

or by mioimizing lhe relative collision velNity bctweenobjecis in the geosynchronous region. To effeatively eliminatethe satellite mass, th€ satellite must be removed from Eanh

orbit; this is not operationally practical since it requires a deltavel$ily of more than I kny'!€c. Without station keeping, therelative collision velocity of objects in geostationary orbit willincrcase to an avemge of 0.5 km/sec. There is an orbit at geo-

synchronous altilude where much k)wer vcl(rities can effec-tively be accomplished for uncontrolled satellites. The orbithas b€€n referred to as "the stable plane orbit" (Friesen. cl al..r 992-A).

Us€ of thc stsble pbnc for enylroDncnt manegementAt geosynchronous altilude, the precession or "wobble" of theorbilal plane occurs about a plane which ilt inclined 7.3degrees to the Eanh's equator. It is this prec'ession which pro-

duces orbital paths which differ by as much il,i 14.6 degrces.and prdluce collision velocities as high as 0.8 km/sec. If asatellite is orbired in this "stable plane," ir would have anorbital inclination of 7.J degrees. and a right ascension ofascending nodc of zero degaees. and would not have any wob-ble. That is, without station keeping, all objccts in lhe stableplane orbit will always be moving in the same direction. solhat if collisions (L-cur. the relative vel(xity will b€ verysmall..,less than 0.(n5 knl/sec. Satellites can easilt be con-

34

structcd to avoid fragmenting at this low collision velocity.Conscquently, collisions benreen spacecraft are nor likety toproduce any fragments large enough to b,rea-kup arDlher space-craft, and collisional cascruling is not possible.

Th€ sr.able planc orbit is not a geostalionary oabit. That is,from the groun4 a lratellire in rhis orbit will move 7.3 degreesNonh and Sourh of the equator. For gmund antcnnas wilhoutNonh-South trrcking, or altennas which require a high signalstrength, this may not be a desiiabie orbit, For those groundstations with Nonh-South trackilg, it can be a highly desirableorbit, since it rcquires or y 5% of the station keeping fuel of ageoslationary o.bit. Many users bave already adopted the prac-tic€ of not u6ing North-South station keeping in order toextend the satellite lifc; however, until recently, these userswer€ prcvented from launching into the stable plale b€causeof a ruling by the Intemarional Frequency Regisration Boadwhich limited sarcllite inclinatioos to less than 5 degrees. lnMarch. 1991. this limitarion was rescinded. Consequently, inth€ futur€, both the stable plane orbit and geoshrionary orbitwill have us€rs driven by oconomic considera(ions. Th€rcfore.environment management of the geosynchronous rcgion ne€dsto consider both types of orbits.

Fmm an environmental management perspective, use ofboth the stable plane and the geostationary orbit is preferred tousing only the geostationary orbit. Use of geostationary orbitalone, without station keeping, leads to highercollision veloci-ties than using bo{h. From arl operational p€rspective, if bothare used, it may be desirable to require rhe us€r of one of th€orbits to maintain a slight eccentricity so that the two orbitalpalhs cannot intersect. Howeve., this should not be necessaryfor satellites which do not rnaintain station kceping since thecollision probabilities are no different than for any uncon-trolled saiellites at geosyDchronous altitudes.

If the stable plane is used fo. gcosynchronous operations,then the use of a near-by grave-yard orbit becomes more prac-tical. Objects could be placed only a lew hundred kilometenabove the geosynchronous stable plarrc, and still be very nea{,or in, a grave-yard stable plane which is inclined slightly moiethan 7.3 degrers. This means rhat collision velociries in thegrave-yard orbit would also be less than 0.005 kn/sec- so thatil a collision (xcun€d, the debris would not spread ro geosyn-chtonous altitude. However, if an object is originally launchedrnto geostationary orbit, the delta velocity required to changeto a stable geosynchronous or saable grave-yard oIbit is pro-hibitively high...nearly,lOO m/sec.

A linal option of the srable plane is to use rhc rwo srablc

Fnnls at geosynchmnous allitude located ove. 75 delrees Eastand lO5 degrees Wesl as a grave,yard orbit. These Iwo pointsare considered desirable operational ltrcalions because Fasl-WesI station keeping is noi requiaed. However, without properenvironmcnt management, thes€ locations would suffer thehiShesr orbilal debris flux. The tendency of objects to movetoward these two points make them an even more stable grave-yard location than any other location in the stable plane or in a

Donald J. Kessler

higher grave yard orbit. Collision velocities at the stabl€ pointswould approach zero, and ifany object had a collision velocitygreater than zem, any collisions would damp out relativemotion uniil the object came to rest at a stable point The moremass placed in th€se stable points. the more stable rheybecome. Consequently, they could rep.€sem a long term solu-tron to managcnr€nt of the orbital debris in seosvnchronousorbit.

Concludlng rsDrrks coocerning gcoqrnchronous ortitAn ad€quaE environm€ntal maoagemc[t strategy does norexist for orbital dcbris in geosynchronous ortit. The use ofgrave-yad orbits dtxs not addrcss the mo.e serious shon termsources of debris: the accidental explosions of upper stagesard stored energy devices on satellites. Neither do these pro_posals significaltly reduce the hazard caused by the long termsources of colliltional fragmentation.

Cun€nI operational practices in oa near geosynchmnousaltitudes combined wirh the long orbital life of debris gener-ated as a result of fhese operaiions make an environmenmlmanagement strdtegy desirable. Some geosynchronous opera-lors are unilaterally p€rfbrming maneuvers in the belief thatihey are contributing to proper environment management.With ,€ss operdtional expense, these operator might make amuch larger contribution to environment manaqemenl. once asrategy has been cstablished.

The cudeot hazad io spacecraft in geosynchronous orbitfrom orbital debris is low and is likely smallcr than the hazardfrom natural meteoroids. However, luture activities in the geo-synchronous .egion may be on a scale much different thantoday's operatioos. We would be ill advisql to preclude theseopemtions because of poor envircnment management pmcticesof today. Rcquiring ope.ators to deplete excess fuet in upperstages left in geosynckonor.rs orbit would be a much moreeffeclive managemenl practlce lhan requiring operalon toman€uvgr to a gBve-yard orbit.

Other options to manage orbital debris irt geosynchronousorbit should be considered. hiorities based on the rrade-offtlelween op€rational exFrense$ and an effective environmenlmanagement str-ategy should be established_ ln order to do this.better mrdels need ro be developed. These modelli should bebased on better data obtained fronr ground lests of sarellitebreakups. and the models should be validated wilh berrerobs€rvational data of the environment in geosynchronousorbil. Until an environmental management straregy is estab-lished which considers the cost effectiveness of all oDrions. ilis premalure to establish policy adoptilg one option overanomer.

REFERENCES

Badhw&. C.D. and P.D. Anz-M€ador (A). .Mass tlslimarion in $cBr€atups of Soviel Sar.llires.' .loumal of Ttte Brilish lnrerDtaneraryS€iery. vol . , {1. r99O

The current and future environmenl

Brdbwar,0.D.. A. Tnn. and R.C. Rcynolds (B). 'velocirl Pcnurbarion

Disldbutiofls in tha Brcrtup of Aniicial Sar.llir.s," Jqrr|lal of

Speecrrft and Roclcts. Vol. 27, No. 3, I yrrr.

E,frett. R.A.. R.P Bemhlrd, and D.S. McKay. "lmp..r Hol€s and lmpacr

Flur on Retuned Solar Mar t uvcr Mar.rid." Lunar ard Plancury

Scierre XiX. March l4-18. 1988.

8crohard, R., W. Dnvidson. aod F. Hirr.. "Preliminary Analysis of LDEF

Instlrn€nl ,40187 l, Location Al l.R, PrEscnrcd at znd LDEF Pos!

Retnevd Symposium. Jun€, 1992.

B.ss. T.D. Mass Disribution of OrbitinS Man-Mad€ Space tlb.is.'

NASA TND D,E I08. I9?5,

Brooki D.R., C.C. Cibson, and T.D. Bess. "htdicrio8 rhe babiliry thar

franh-OrbitinS SP3c€(Taft Will Collide witl| Mm-Madc Objecrs in

Spdcc. Space R.scu€ a'd Safety. pp. 79_l19. 1975

Bnrk. R. PeMnal communicarion, GDODSS site. Maui. 1992.

Chot'olo!. V.A.'Disposal of Spacecrafi at End of Life in CJeosynchmnous

Cxr i r . 'Vol 27. No.4. 1990.

Chnsuans€n. E-. ,.R. Hom. dnd J.L. Crews. 'Augmcnrsrion of Crrbilsl

Dekis Shieldio8 for Space Starion FrEedom. AIAA 9G1665. l9q).

Chnstiansen. E. NASA. JSC Dersonal communicatron- 1992.

Donahoo. M.E. "Collision Prob.biliry of Future Manned Missions wilh

ObJects in Eanh frrir. MSC lnremal Nore No.70-FM 168. Oct.*rcr,

1970

Echler. P. rnd D. Rer. 'Chain Reaction\. AIAA Paper 90 1165 p.e

sented at AIAA/NASA/DOD frital Debns ConferEnce. 1990

Fannella, P. and A. Cord€lli. The Pmlifer.don of mrial Fragments: A

Simple Malhema(ical Model. Sljerue & Clobal Secu.ity, Vol. 2,

pp. 365-.178, 1991.

Fnesen. L.J.. A.A.lackson. H.A. Z,$k. "rd D., Kes.lcr (A). "Analyns of(hrtal P€nurb3xons Acting on Obie($ in frits n€ar CEO,' .lournal

ol Ceophy srcal Researfh, Vol. 97, No E3. pp. 3845-1861. I q92.

Fri€sen. L.J.. A.A. Jdckson, H.A. Zook. aM D.J. Kcsslcr (B). Rcsulrs in

Cjrbital r-volution of Obiects in the Gcosynchmnous ReSion ' Joumal

otcuidnce. Conirol. and Dynarnics. Vol. 15. No. I, 1992.

Flury. W.'tollision Probability and Spa.€r:rafi Dispositiofl in lhe

Geostat ionrJ- Orbr l . Adv Space Rcs. . Vol l l .No. l : . lSql .

Goldst€in, R. and t-. Randolph. "Rings of E3nh DelecGd by Orbiral

Debfi s Ra&r, JPL Progrcss Repon 42- 1 0 1. Mry. l 5 l !rn}.

Guemnnprcz. V [oping wi(h Geosrationary OvcrcmwdinS, 5th

NASA/ESA C(xrdination Meeiing, (ktober, I990.

Griin. E. H.A. Z{Dk, H. Fe!hti8, and R H. Gicse. alollisional Ealsncf of

the Mc|eorrric Conrplcr. lcds. Vol 6:. pp 244. 1985.

Hechler, M. and J C. van drr Ha "Thc t.robabifity of Collisions on fir

Ceostation!ry Ring,' ESA ,oumal. vol.4, No. l. 1980

Hechter. M. tollision Probabilily ar Ceosynchronous Allitud€s.' Adv

Spnce Res. . Vol .5. No. l . 1985.

Henire. H dnd J. Sranl€y. "Optical Obs€ftations of Orbiral tt€t'ris,'

AIAA paper AIAA-9(}-ll40 presented at AIAA./NASA/DOD Orbilal

Debns Confcreree. l9q).

Horl. F. R.P Bemhard. J.L Warrcn. T.H. Sfa. D.E Brownlee.

M.R. Laurance. S. McssenSer. and R.B. Petenon "PrEliminary

Anllysis of l-DEIr Insrurient A0lE7-l: The Chcmisfy of

Micromeicomids Expenincnr. NASACP,.ll.H. |q)l.

Johnson. N.L. ard D.J. Naocr. History of On-frir Satcllire

Frirgrnenurrons.'Ielcdync llmwn Engine€rin8 Technical Repoi

CSgl TR JS(' fi)8. July. l99l

Johnson. N.t-. Reported In Avianon Wetk. March q. 1991.

Kcssler. I).J 'Avcrsge Relarive Velociry of Sporadic Meteoroids in

35

Io.erplaneirry Space, AIAA Joumal, Vol7, No. t2. D€lember. t969.K.sslcr, D.J. tollision FEqu€ncy of A.tificial S{r.llircs: Thc Crcadon of

r Dcbfis Bch," Joumal of C€ophysical Rcs€€rch, Vot_ 83, No. A6,pp. 2637-2646. Jun€, l9?8.

Kcssl6. D.J. (A). "Sourccs of (hitd D€brtu and rhe Proj€cLdEovironrEnt for Futurc Sp&.ctnfi." Joumal of Spacecraft and Rockels.

Vol . 18, No.4, pp 157-160. 1961.

Kcsslcr. D.J. (B). "Dcrivation of lhc Collision hob.bility B€lwc€n(Htting Objc€lsr Th€ Lifeti:nes of Jupitcr's Ourcr Moons. lcarus.Vol.48. pp. 19-48. l9El.

Kcssler. D.J.. R.C. Reynolds. and P.D. An?,Meador "O'birat DcbnsEnvironrncnr for Spoctcrafi D€siSrFd io Opcr|te in Low Eanh Odtir,.NASA TM'1fi}47 I, April, 1989.

Kcssler, D.J. (B). "Technical lssues Aslocisled wi& ftid D€bris inGEO." ftitd Debris Monibr. Vol. 2. No.4, 1989.

Kcsslet. DJ.'tollision Prcbabihy ar Low Alrirudes Resutting fromBlliptical Orbits.'Adv. Spacc Res., Vol. 10, No. ]-4. pp. 193-396.l9q).

Kessler. D.r. (A). Collisional Cascading: The Limils of Popularion

Growrh in bw E3ni m'll" Adv. in Spac€ Res.. Vot I l. No_ t2.pp.63-66. | 99 I

Kessler. D.J. (B). Orbiral Dckis Environmenl for Soacccrafi an lrw

tirih fx)rt, l mal of Spacecmfi nnd RockeLs. Vol. 28, No. L pp

347-351. t99t .

Kcssler. D.J. 'Origrn of ftriul De.bris Inpacrs on Long h,r ionErposurE Ficilhy's (LITEF) Tniling Surfa€es, Prescmed nt 2nd LDEFPoiil Relri€vrl Symposium. Jun€, lql2.

L-ec. J., D Budcn, T. Albcn. W. Margopolous, J. Angclo, snd S. tapin.'"Technology Requiremen$ for the Disposal of Spa€e Nuctear Power

Sourccs aod lmplicaiions for Space D€bris Managenenr Strare8ies."

AIAA yaFr AIAA-q)-1368. presenled at AIAA,/NASA,/DOD Orbir4l

Debris Conf€rence. 1990.

lnftus. J.P.. D.J Kessler. 3nd P.D. Anz-Mcador "Manage.nenr of rhe

ffital Debris Envircnm€n!' Acta Asrooaulica- Vol. ?6. No 7.pp 17 t486. t992.

Mcxnight. D.S. Examinarion of Possibl€ Collisiors in Space," tAF papcr

lAA,9r-593. l9St.

Mueller, A.C. and D.J- Kessler. "fhe Eff€c$ of Paniculates f.om Sohd

Roskcl Motors Fir€d in Spac€." Adv. Space Rcs-. Vol. 5. No. 2, t985.

Mulholland. J.D.. S.F. Singer. J.P. Oliver. J.L. W€in8erg. WJ. Cooke.

N.L. Montague. J.J. Wonman, P.C. Krss€I. and W.H. Kinad. .tDE

Spati(}Temporal lmpact nuxes and HiSh Tinr-Rexolurion Srudies ofMuhilmpacr Evenls and t-ong-Livcd Debris Clouds. NASA CR-3l14- 199t .

Perel. L. Physical Nature and T€chnical Air.ibures of lhc CeoslarionanOrbil. UN Committoe on the Pea.tful Utes of Outer Spdce. t978.

Reynolds, R.C. "Review of Cu.rcnr Activirics io Model and Mfasurc d|€(Xrilal D€kis Envircnmcnl in f-ow-Eanh frrr," Adv. SDace Res..Vol. 10. No. L pp. 159-371, 1990.

Royal Aircraf. Esrablishrnenr. Th€ RAI Table of &nh Saretlires

t98:r- t985. ' r9E6.

Slansb€ry. E.C. and C.C. Pilts. "Radar Mcasur€menb of rhe (hitat

Debris Eovrronmenl, ESA SP,342, October, 1991.

Sla$bery. E.C. Charr.leriation of th€ Ort ilal Deb.is Environmcnl

UsinS rhe Haystack RadaJ. ' JSC-3221l. April 24. 1992.

Suddclh. DH-, Debns in the ceosradonary (h'it Ring: Thc EndlersShooling Gallery' - Th€ Nc.csshy for a Disposal Policy. NASA

cR-:160, t985.

36

Tafi. L.C.. D.E. 8e{ny, A.J_ yakutis .nd p.M.S. Randalt. ..l,(}* Aliirude.On Centirncr4r. Space D€bris Se.rch al Lincoln f"rborarory's (M.l.T.)

ErlErimcnlal Test Sysr€m . Adv. Sp.ce Res.. Vot. 5. No. 2. Fp t5-45.t9E5.

Talcnl. D.L. "Anaiydc Modcl for (hit tlbris El|vironmcdrMaragcrncnt." Jo{rnnl ol SFcecrnft ad Rocket{, Vol. 29, No. 4.pp. :O8-5 ll. July-A09., 1992.

T.dcscbi, W.. J. ConrEll, D. McKriShr, F. Alladadi, A. Rcinh.rdt,R. Hunl, ard D. Hogg. "D€rcrmining rhc Effc.rs of Spec Dcbris

Donald J. Kessler

lrpacts on Spacecrafi Stnrclures. ' I AA paD.r. IAA 9 | 594. | 99 ITlbmp{on. T.W.. R.M. Coldsrcin. D.B. Cdmpbcll. E.C. Sransbcr}. amt

A.E. Poner. 'Radar Dclccrion of Ccnrirhcrer-Sized Orbiui Debris:Prcliminary Arccibo Otrlicrvarions ar t2.5 cm Wavetcngft.'G.ophysical Rcscarh t€Iers. Vol. 19. No. l. lqrl

Z.d*. H.A.. D.S. McKay, and R.P. Bemhard "Resutrs liom RcrumcdSpa€lcraft Surta(:cs, A|AA paper AAIA-9G I 149. A|A.4/NASA./DOI)

Orbit i Debris ConfcrE.Lc. l9S0

izook. H.A. NASA.JSC personal communicarion. t992.